EP3553185A1 - Method for acquiring information on prognosis of breast cancer, and device for determining prognosis of breast cancer - Google Patents

Abstract

Disclosed is a method for acquiring information on prognosis of breast cancer, the method comprising: a measuring step of measuring expression levels of genes of IL6ST, NPY1R, ELOVL5, OPA1, ASAH1, ALDH6A1 and SYBU in an RNA sample prepared from a biological sample collected from a subject; an analyzing step of analyzing the measured expression levels of genes; and an acquiring step of acquiring information on prognosis of breast cancer based on an analysis result.

Description

TECHNICAL FIELD
• The present invention relates to a method for acquiring information on prognosis of breast cancer. The present invention also relates to a device for determining prognosis of breast cancer.
BACKGROUND
• As multigene assays for recurrence prediction aiming for tailored treatment of breast cancer, Oncotype DX (registered trademark) (Genomic Health, Inc.), MammaPrint (registered trademark) (Agendia), PAM50-ROR (Nanostring Technologies, Inc.), EndoPredict (registered trademark) (Myriad Genetics, Inc.), Curebest (registered trademark) 95GC Breast (Osaka University and Sysmex Corporation: see US Patent Application Publication No. 2011/0263444 ), and the like are known. These multigene assays are excellent in predicting early recurrence of breast cancer within 5 years after surgery.
• As postoperative adjuvant therapy for hormone receptor positive breast cancer, it is standard to administer hormones such as tamoxifen for 5 years after surgery. However, breast cancer may recur also more than 5 years after surgery. Especially, in hormone receptor positive breast cancer, late recurrence more than 5 years after surgery tends to be higher than that in negative breast cancer. In recent years, it has been shown that hormone therapy which administers hormones for 10 years after surgery (extended endocrine treatment) has an effect of reducing risk of recurrence and death, by large-scale clinical trials aTTom and ATLAS. However, long-term administration of hormones has problems of side effects. Also, when extended endocrine treatment is performed for all breast cancer patients, another problem of an increase in medical expenses also arises. Therefore, in clinical sites, a need for a method to predict late recurrence more than 5 years after surgery has rapidly increased in order to decide patients to which extended endocrine treatment is applied. Therefore, attempts have been made to predict late recurrence by the above multigene assays.
SUMMARY OF THE INVENTION
• However, the above multigene assays are methods suitable for predicting early recurrence within 5 years after surgery. Early recurrent breast cancer has high cell growth ability, and late recurrent breast cancer has low cell growth ability. Thus, since biological characteristics of early recurrent breast cancer and late recurrent breast cancer differ, prediction accuracy of late recurrence by the multigene assays is limited. Accordingly, it is desirable to develop a method to enable prediction of late recurrence of breast cancer with high accuracy.
• The present invention provides a method for acquiring information on prognosis of breast cancer, the method including: a measuring step of measuring expression levels of genes of IL6ST, NPY1R, ELOVL5, OPA1, ASAH1, ALDH6A1 and SYBU in an RNA sample prepared from a biological sample collected from a subject; an analyzing step of analyzing the measured expression levels of genes; and an acquiring step of acquiring information on prognosis of breast cancer based on an analysis result.
• The present invention provides a device for determining prognosis of breast cancer, the device including a processor, and a computer containing a memory under control of the processor, wherein the memory is recorded with a computer program for causing the computer to execute the steps of: acquiring information on expression levels of genes of IL6ST, NPY1R, ELOVL5, OPA1, ASAH1, ALDH6A1 and SYBU in an RNA sample prepared from a biological sample collected from a subject; determining prognosis of breast cancer based on the information on the expression levels of genes; and outputting a determination result.
• The present invention provides a computer program for determining prognosis of breast cancer recorded on a computer readable medium, the computer program causing a computer to execute the steps of: acquiring information on expression levels of genes of IL6ST, NPY1R, ELOVL5, OPA1, ASAH1, ALDH6A1 and SYBU in an RNA sample prepared from a biological sample collected from a subject; determining prognosis of breast cancer based on the information on the expression levels of genes; and outputting a determination result.
• The present invention enables prediction of prognosis of breast cancer in a subject, especially, late recurrence after surgery with high accuracy.
BRIEF DESCRIPTION OF THE DRAWINGS
• Fig. 1 is a decision tree of a treatment policy for patients with estrogen receptor (ER)-positive, locally advanced breast cancer.
• Fig. 2 is a decision tree of a treatment policy for patients with ER-positive early breast cancer (pN0: no lymph node metastasis).
• Fig. 3 is a decision tree of a treatment policy for patients with ER-positive early breast cancer (pN1-3: with lymph node metastasis).
• Fig. 4 is a decision tree of a treatment policy for patients with ER-positive early breast cancer (pN1: with lymph node metastasis).
• Fig. 5 is a schematic diagram showing an example of a device for determining prognosis of breast cancer.
• Fig. 6 is a block diagram showing a hardware configuration of the device for determining prognosis of breast cancer.
• Fig. 7 is a flowchart for determination using the device for determining prognosis of breast cancer.
• Fig. 8 is a graph showing a relationship between the number of probe sets and accuracy or negative predictive value (NPV).
• Fig. 9A is a graph showing a relationship between period after surgery and distant recurrence-free survival rate in a training set.
• Fig. 9B is a graph showing a relationship between period after surgery and distant recurrence-free survival rate in a validation set.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
• A method for acquiring information on prognosis of breast cancer of the present embodiment (hereinafter, also simply referred to as "method") enables acquisition of information on prognosis of breast cancer of a subject, by measuring and analyzing expression levels of predetermined genes. As the information on prognosis of breast cancer, information on risk of recurrence after surgery, especially, information on risk of late recurrence after surgery can be acquired. More specifically, according to the method of the present embodiment, the risk of recurrence after surgery in a subject, especially, whether the risk of late recurrence after surgery is high or low can be accurately predicted. In the present specification, the term "late recurrence" means recurrence more than 5 years after surgery. The phrase "more than 5 years after surgery" refers to a day after 5 years have elapsed from a day of surgery (that is, it does not include a day of 5 years after surgery). The method of the present embodiment is especially suitable for predicting risk of recurrence more than 5 years within 10 years after surgery.
• By providing information obtained by the method of the present embodiment to a person who decides diagnosis and treatment policy (for example, a doctor or the like), it is possible to assist diagnosis of recurrence risk of breast cancer. Based on the information obtained by the method of the present embodiment, it is possible to decide necessity of hormone therapy over a long period of 10 years, so that reduction of side effects and suppression of medical expenses are expected.
• In the method of the present embodiment, first, expression levels of genes of IL6ST, NPY1R, ELOVL5, OPA1, ASAH1, ALDH6A1 and SYBU in an RNA sample prepared from a biological sample collected from a subject are measured. Hereinafter, these seven genes are also referred to as "first gene group".
• In a preferred embodiment, in addition to the above seven genes, an expression level of at least one gene selected from CALB2, ASAP2, RAB5C, PTP4A2, ABCC10, OXTR, HSPA2, SMURF2, SLC7A8, RALA, ADRA2A, MYCBP, CX3CR1, ERCC1, DNAJA3, NINJ1, C4orf43, IFI35, ZNF688, SNX1, CREBL2, HPN, NME3, STS, KLF7, PDHB, NKX3-1, DEXI, GSTM3 and LCMT1 is further measured. Hereinafter, these 30 genes are also referred to as "second gene group". In the present embodiment, it is particularly preferable to measure the expression levels of all 37 genes contained in the first and second gene groups.
• The subject is a breast cancer patient. It is known that ER-positive breast cancer patients tend to recur more than 5 years after surgery, compared with ER-negative breast cancer patients. Accordingly, an ER-positive breast cancer patient is preferable as the breast cancer patient. The subject may be undergoing preoperative adjuvant therapy as long as a biological sample containing breast cancer cells can be collected. Examples of the preoperative adjuvant therapy include administration of hormones, administration of anticancer agents, and combinations thereof. Examples of the hormones include tamoxifen, toremifene, leuprorelin acetate, goserelin acetate, anastrozole, fulvestrant, and the like. Examples of the anticancer agents include adriamycin, epirubicin, paclitaxel, docetaxel, cyclophosphamide, methotrexate, fluorouracil, and the like.
• Since the method of the present embodiment can predict recurrence risk more than 5 years after surgery, the subject may be a subject who did not have recurrence of breast cancer, for example, for 5 years after surgery. The subject may be a subject whose risk of early recurrence of breast cancer has been determined to be low by a prognosis prediction method publicly known in the art. As such a prognosis prediction method, a method capable of predicting early recurrence of breast cancer is preferable. Examples include multigene assays such as Oncotype DX (registered trademark), MammaPrint (registered trademark), PAM50-ROR, EndoPredict (registered trademark) and Curebest (registered trademark) 95GC Breast. The term "early recurrence" refers to recurrence within 5 years after surgery. In the present specification, the phrases "within 5 years after surgery" and "for 5 years after surgery" refer to a day until 5 years have elapsed from the day of surgery (that is, they include the day of 5 years after surgery).
• As described above, ER-positive breast cancer tends to recur late, and ER-negative breast cancer tends to recur early. One of causes of breast cancer recurrence includes micrometastasis. The present inventors have searched for marker genes for the purpose of establishing an inspection method for predicting recurrence time, when breast cancer remains as a micrometastasis in the body after surgery. Specifically, patients with breast cancer recurrence have been divided into an early recurrence group and a late recurrence group according to recurrence time, and differences in gene expression levels between the two groups have been comprehensively analyzed. As a result, the present inventors have identified the above 37 genes as genes associated with late recurrence of breast cancer. Thus, in a preferred embodiment, the subject is an ER-positive breast cancer patient, and expression levels of genes of IL6ST, NPY1R, ELOVL5, OPA1, ASAH1, ALDH6A1, SYBU, CALB2, ASAP2, RAB5C, PTP4A2, ABCC10, OXTR, HSPA2, SMURF2, SLC7A8, RALA, ADRA2A, MYCBP, CX3CR1, ERCC1, DNAJA3, NINJ1, C4orf43, IFI35, ZNF688, SNX1, CREBL2, HPN, NME3, STS, KLF7, PDHB, NKX3-1, DEXI, GSTM3 and LCMT1 in an RNA sample prepared from a biological sample collected from the subject are measured.
• In general, colorectal cancer, lung cancer, ER-negative breast cancer and the like are known to have a fast growth rate among cancers. When such cancer with a fast growth rate exists in the body as a micrometastasis which cannot be pointed out by image inspection, it is considered that the cancer will recur within 5 years after surgery. However, no one knows whether or not there is such a micrometastasis at the time of surgery. While cure may be achieved by effects of surgery, preoperative or postoperative drug therapy and the like, the therapeutic effects cannot be examined even by assays such as microarray. Therefore, in clinical practice, when a patient has no recurrence for 5 years, it is common to finish follow-up of the patient, considering that cancer has been cured. However, ER-positive breast cancer has a slower growth rate than ER-negative breast cancer, and since hormone therapy is effective, cases recurring more than 5 years after surgery are seen. Therefore, it is dangerous to consider that cancer has been cured in ER-positive breast cancer even when the patient has no recurrence for 5 years after surgery. On the other hand, there are not a few cases of ER-positive breast cancer recurred within 5 years after surgery. The present inventors have considered that, among ER-positive breast cancers, there exists breast cancer with a fast growth rate that recurs within 5 years after surgery when there is micrometastasis at the time of surgery by about 30 to 50%, as with ER-negative breast cancer. In a case where such breast cancer with a fast growth rate and breast cancer with a slow growth rate can be accurately found in an inspection, it can be considered that breast cancer has been cured when ER-positive breast cancer with a fast growth rate does not recur for 5 years after surgery. When breast cancer is cured, administration of hormones more than 5 years after surgery is unnecessary. In the present embodiment, the expression levels of all the above 37 genes in an RNA sample prepared from a biological sample collected from an ER-positive breast cancer patient are measured and analyzed, whereby risk of late recurrence of the patient can be accurately predicted. When ER-positive breast cancer does not recur for 5 years after surgery in a patient whose risk of late recurrence has been determined to be low, extended endocrine treatment can be safely omitted.
• The biological sample is not particularly limited as long as it contains breast cancer cells of the subject. Examples thereof include cells collected by fine-needle aspiration, tissues collected by surgery or biopsy, and the like. A formalin-fixed paraffin-embedded (FFPE) specimen prepared from cells or tissues collected from the subject may be used as the biological sample. In a preferred embodiment, the biological sample is a cell or tissue collected from a subject not yet undergoing treatment for breast cancer, or a breast cancer patient undergoing only preoperative adjuvant therapy with hormones, or an FFPE specimen thereof.
• The RNA sample contains RNA extracted from the biological sample or a nucleic acid derived from the RNA, and is a sample to be subjected to measurement of the gene expression levels. A method for extracting RNA from the biological sample is publicly known per se. When the biological sample is a cell or a tissue, extraction of RNA can be performed, for example, as follows. First, the biological sample is mixed with a solubilizing solution containing guanidine thiocyanate and a surfactant. Physical treatment (stirring, homogenizing, ultrasonic disruption, etc.) is performed on the resulting mixed solution to release RNA contained in the biological sample into the mixed solution. Thus, RNA can be extracted from the biological sample. The extracted RNA may be purified. For example, RNA can be purified by centrifuging a mixed solution containing RNA to collect a supernatant, and extracting the supernatant with phenol/chloroform. Extraction and purification of RNA from the biological sample may be performed using a commercially available RNA extraction kit.
• When the biological sample is an FFPE specimen, extraction of RNA can be performed, for example, as follows. First, xylene is added to the FFPE specimen and deparaffinization treatment is performed. The deparaffinized specimen is immersed in ethanol to make it hydrophilic. By treating the hydrophilized specimen with a protease to release formalin-crosslinked nucleic acid, RNA can be extracted from the specimen. Extraction and purification of RNA from the FFPE specimen may be performed by using a commercially available nucleic acid extraction kit for FFPE specimen.
• In the present embodiment, RNA extracted from the biological sample and purified may be used as the RNA sample. Alternatively, cDNA obtained by reverse transcribing mRNA contained in the RNA may be used, or cRNA obtained by in vitro transcription (IVT) amplification of the cDNA may be used as the RNA sample. A method for acquiring cDNA and cRNA is publicly known per se.
• In the present specification, the term "expression level of a gene" means the amount of mRNA transcribed from a gene or the amount of substance reflecting the amount of mRNA. Examples of the substance reflecting the amount of mRNA include cDNA, cRNA, and the like. Since the amount of mRNA transcribed from a gene is often very small, it is preferable to measure the amount of cDNA or cRNA. The expression level of a gene may be represented by either concentration, copy number, or a measured value indicating concentration or copy number.
• The nucleotide sequence of each of the above genes is publicly known per se. Information on these nucleotide sequences can be obtained, for example, from a publicly known database such as a database (http://www.ncbi.nlm.nih.gov/) provided by the National Center for Biotechnology Information (NCBI) of the United States National Library of Medicine. Table 1-1 and Table 1-2 show UniGene IDs, GenBank accession numbers and probe set IDs of each gene. The probe set IDs shown in Table 1-1 and Table 1-2 are numbers for specifying a probe set loaded on GeneChip (registered trademark) Human Genome U133 Plus 2.0 Array (Affymetrix, Inc.). Each probe set contains 11 to 20 probes. Information on nucleotide sequences of probes included in a probe set specified by each probe set ID can be obtained from Affymetrix's web page (http://www.affymetrix.com/analysis/index.affx) and the like. The "UniGene ID" is ID number of UniGene which is a database published by NCBI. The "GenBank accession number" is an accession number of public database GenBank used for designing a sequence of each probe of microarray GeneChip (registered trademark) manufactured by Affymetrix, Inc. The "gene number" is a number assigned to a gene corresponding to each probe set ID.
[Table 1]
• Table 1-1

  Gene number	Probe set ID	Gene symbol	UniGene ID	GenBank accession number
  1	212196_at	IL6ST	Hs.532082	AW242916
  2	205440_s_at	NPY1R	Hs.519057	NM_000909
  3	208788_at	ELOVL5	Hs.520189	AL136939
  4	214306_at	OPA1	Hs.594504	AA209332
  5	212195_at	IL6ST	Hs.532082	AL049265
  6	210980_s_at	ASAH1	Hs.527412	U47674
  7	204864_s_at	IL6ST	Hs.532082	NM_002184
  8	221590_s_at	ALDH6A1	Hs.293970	AF130089
  9	218692_at	SYBU	Hs.390738	NM_017786

[Table 2]
• Table 1-2

  Gene number	Probe set ID	Gene symbol	UniGene ID	GenBank accession number
  10	205428_s_at	CALB2	Hs.106857	NM_001740
  11	206414_s_at	ASAP2	Hs.555902	NM_003887
  12	201156_s_at	RAB5C	Hs.650382	AF141304
  13	216988_s_at	PTP4A2	Hs.470477	L48722
  14	213485_s_at	ABCC10	Hs.55879	AK000002
  15	206825_at	OXTR	Hs.2820	NM 000916
  16	211538_s_at	HSPA2	Hs.432648	U56725
  17	205596_s_at	SMURF2	Hs.515011	AY014180
  18	202752_x_at	SLC7A8	Hs.596643	NM_012244
  19	214435_x_at	RALA	Hs.6906	NM_005402
  20	209869_at	ADRA2A	Hs.249159	AF284095
  21	203360_s_at	MYCBP	Hs.591506	D50692
  22	205898_at	CX3CR1	Hs.78913	U20350
  23	203720_s_at	ERCC1	Hs.435981	NM_001983
  24	205963_s_at	DNAJA3	Hs.459779	NM_005147
  25	211000_s_at	IL6ST	Hs.532082	AB015706
  26	203045_at	NINJ1	Hs.494457	NM_004148
  27	218513_at	C4orf43	Hs.267446	NM_018352
  28	209417_s_at	IFI35	Hs.632258	BC001356
  29	213527_s_at	ZNF688	Hs.301463	AI095896
  30	201716_at	SNX1	Hs.188634	NM_003099
  31	201988_s_at	CREBL2	Hs.591156	BF438056
  32	204934_s_at	HPN	Hs.182385	NM_002151
  33	204862_s_at	NME3	Hs.514065	NM_002513
  34	203769_s_at	STS	Hs.522578	NM_000351
  35	204334_at	KLF7	Hs.471221	AA488672
  36	208911_s_at	PDHB	Hs.161357	M34055
  37	204863_s_at	IL6ST	Hs.532082	BE856546
  38	209706_at	NKX3-1	Hs.55999	AF247704
  39	203733_at	DEXI	Hs.592051	NM_014015
  40	213702_x_at	ASAH1	Hs.527412	AI934569
  41	202554_s_at	GSTM3	Hs.2006	AL527430
  42	221515_s_at	LCMT1	Hs.337730	BC001214

• A method for measuring the expression levels of genes is not particularly limited, and can be appropriately selected from publicly known measurement methods. Examples of the method include microarrays, quantitative RT-PCR, quantitative PCR, and the like. In the present embodiment, it is preferable to measure the expression levels of genes using a microarray. In this case, it is more preferable to measure expression levels of genes using probe sets specified by the probe set IDs shown in Table 1-1 and Table 1-2. For example, when the expression levels of seven genes of the first gene group are measured, it is preferable to use probe sets specified by nine probe set IDs shown in Table 1-1. When the expression level of at least one gene selected from the second gene group is further measured in addition to the above seven genes, it is preferable to use the probe sets specified by nine probe set IDs shown in Table 1-1 and a probe set specified by a probe set ID corresponding to the gene to be measured shown in Table 1-2. When the expression levels of all 37 genes contained in the first and second gene groups are measured, it is preferable to use the probe sets specified by 42 probe set IDs shown in Table 1-1 and Table 1-2.
• The microarray is not particularly limited as long as it is a chip in which a probe capable of specifically hybridizing to mRNA transcribed from each of the above genes, or cDNA or cRNA derived from the mRNA (hereinafter also referred to as "target nucleic acid") is immobilized on a suitable substrate. The probe can be appropriately designed based on the nucleotide sequence of each of the above genes. The length of the probe may be 10 to 50 nucleotides. The microarray itself can be prepared by a publicly known method. In the present embodiment, it is preferable to use a microarray loading probe sets specified by the probe set IDs shown in Table 1-1 and Table 1-2. A commercially available microarray may be used.
• As used herein, the phrase "capable of specifically hybridizing" means that a probe can hybridize to a target nucleic acid under stringent conditions. The term "stringent conditions" means conditions commonly used by those skilled in the art under which hybridization of polynucleotides is performed. It is known that the stringency of conditions under which hybridization is performed is a function of temperature, salt concentration of hybridization buffer, probe length, GC content of nucleotide sequence of probe and concentration of chaotropic agent in hybridization buffer, and those skilled in the art can suitably set in consideration of these conditions. As the stringent conditions, conditions described in, for example, Molecular Cloning: A Laboratory Manual (second edition) (Sambrook, J. et al., Cold Spring Harbor Laboratory Press, New York (1989)) and the like can be used.
• When the microarray is used, it is preferable that the target nucleic acid is labeled with a publicly known labeling substance. By labeling the target nucleic acid, it is easy to measure a signal from the probe on the microarray. In the present embodiment, while mRNA extracted from the biological sample may be labeled, it is preferable to label cDNA or cRNA derived from the mRNA. Examples of the labeling substance include fluorescent substances, hapten such as biotin, radioactive substances, and the like. Examples of the fluorescent substance include Cy3, Cy5, FITC, Alexa Fluor (trademark), and the like.
• In measurement by the microarray, the expression level of gene is obtained as a signal from the probe, such as fluorescence intensity, luminescence intensity or current amount. The signal can be detected by a scanner included in a microarray analyzer. Examples of the scanner include GeneChip (registered trademark) Scanner3000 7G (Affymetrix, Inc.), Illumina (registered trademark) BeadArray Reader (Illumina, Inc.), and the like.
• Next, in the method of the present embodiment, the measured gene expression levels are analyzed. In the present embodiment, measurement data (raw data) of gene expression levels may be normalized by a publicly known statistical algorithm for analysis. Examples of such a statistical algorithm include RMA, MAS5, PLIER, and the like. RMA, MAS5 and PLIER can be used, for example, in Affymetrix Expression Console (trademark) software (manufactured by Affymetrix, Inc.).
• The gene expression levels can be analyzed by a publicly known method such as a classification method, a scoring method, or a hierarchical cluster analysis. Examples of the classification method include diagonal linear discriminant analysis (DLDA), between-group analysis (BGA) (see Culhane A.C. et al., Between-group analysis of microarray data, Bioinformatics, 2002, vol. 18, p. 1600-1608), support vector machine (SVM), k nearest neighbor classification (kNN), decision tree, random forest, neural net, and the like.
• Hierarchical cluster analysis can be performed, for example, as follows. Using data on expression levels of a specimen (RNA sample) derived from a subject, data on expression levels in a group of specimens known to have good prognosis, data on expression levels in a group of specimens known to recur late, a distance indicating similarity between specimens is calculated based on the expression levels. Based on this distance, various clusters are formed. Analysis is performed by integrating clusters and creating tree diagrams. Examples of the distance include Spearman rank correlation coefficient, Euclidean distance, and the like. Integration of clusters can be performed, for example, by ward method, furthest neighbor method, distance-between-centroids method, and the like. Among them, the Spearman rank correlation coefficient and the ward method are preferably used.
• Examples of the scoring method include principal component analysis, multiple regression analysis, logistic regression analysis, Partial Least Square, and the like. When the gene expression levels are analyzed using the scoring method, scoring is performed so as to classify a score of a specimen predicted to have a good prognosis and a score of a specimen predicted to have a poor prognosis, based on the expression levels.
• When the gene expression levels are analyzed by DLDA classification method, a discriminant constructed using DLDA can be used. For example, when expression levels of genes are measured using the probe sets specified by the probe set IDs shown in Table 1, the discriminant represented by the following formula (I) is shown as the discriminant. $$\mathrm{D}={\mathrm{\Sigma }}_{\mathrm{i}}\left({\mathrm{w}}_{\mathrm{i}}\times {\mathrm{y}}_{\mathrm{i}}\right)-\left(-\mathrm{4.763756453}\right)$$

  

  wherein i represents a number assigned to each gene shown in Table 2-1 and Table 2-2, w_i represents a weighting factor of gene numbered i shown in Table 2-1 and Table 2-2, y_i represents a standardized expression level of gene according to a formula represented by formula (II): $${\mathrm{y}}_{\mathrm{i}}={\mathrm{x}}_{\mathrm{i}}-{\mathrm{m}}_{\mathrm{i}}$$

  

  wherein x_i represents an expression level of gene numbered i shown in Table 2-1 and Table 2-2, and m_i represents a mean value of expression levels of genes numbered i shown in Table 2-1 and Table 2-2 over specimens,
  and ∑_i represents a sum total over the genes.
[Table 3]
• Table 2-1

  Gene number	Probe set ID	Gene symbol	Weighting factor
  1	212196_at	IL6ST	1.43235997
  2	205440_s_at	NPY1R	0.450736337
  3	208788_at	ELOVL5	1.793041948
  4	214306_at	OPA1	-2.545950461
  5	212195_at	IL6ST	1.539087017
  6	210980_s_at	ASAH1	1.539921375
  7	204864_s_at	IL6ST	1.151743979
  8	221590_s_at	ALDH6A1	1.837908119
  9	218692_at	SYBU	1.149172541

[Table 4]
• Table 2-2

  Gene number	Probe set ID	Gene symbol	Weighting factor
  10	205428_s_at	CALB2	-1.179415269
  11	206414_s_at	ASAP2	-2.323532179
  12	201156_s_at	RAB5C	2.316847761
  13	216988_s_at	PTP4A2	2.002072892
  14	213485_s_at	ABCC10	-2.306701292
  15	206825_at	OXTR	-1.1298499
  16	211538_s_at	HSPA2	0.964294299
  17	205596_s_at	SMURF2	-2.109236485
  18	202752_x_at	SLC7A8	1.363057605
  19	214435_x_at	RALA	-2.288374092
  20	209869_at	ADRA2A	1.377444437
  21	203360_s_at	MYCBP	1.688897659
  22	205898_at	CX3CR1	0.740592577
  23	203720_s_at	ERCC1	1.658860215
  24	205963_s_at	DNAJA3	2.534304027
  25	211000_s_at	IL6ST	0.971312607
  26	203045_at	NINJ1	1.889380552
  27	218513_at	C4orf43	1.429869188
  28	209417_s_at	IFI35	1.323441517
  29	213527_s_at	ZNF688	2.158913805
  30	201716_at	SNX1	2.2492475
  31	201988_s_at	CREBL2	1.921069642
  32	204934_s_at	HPN	1.115597062
  33	204862_s_at	NME3	1.661682832
  34	203769_s_at	STS	-1.392023997
  35	204334_at	KLF7	-1.662199312
  36	208911_s_at	PDHB	2.071027409
  37	204863_s_at	IL6ST	0.825924727
  38	209706_at	NKX3-1	0.611126615
  39	203733_at	DEXI	1.847912036
  40	213702_x_at	ASAH1	1.575733773
  41	202554_s_at	GSTM3	0.648154701
  42	221515_s_at	LCMT1	1.7259299

• m_i in the formula (II) can be acquired as follows. For example, when the method according to the present embodiment is performed on each of a plurality of subjects in one cohort, first, expression levels of genes are measured using the probe sets shown in Table 2-1 and Table 2-2, for specimens (RNA samples) derived from each of the plurality of subjects. From the measured expression levels of genes, a mean value in the specimens of the plurality of subjects is calculated. The obtained mean value of the expression levels of genes can be used as m_i.
• Alternatively, with respect to a plurality of breast cancer patients in the same facility, expression levels of the genes shown in Table 2-1 and Table 2-2 are measured and data are accumulated, and a mean value may be calculated for the expression levels of the genes of those patients. The mean value of the expression levels of the genes thus obtained can be used as m_i in the formula (II) when the method of the present embodiment is performed on a predetermined subject in the facility.
• When the expression levels are analyzed using the discriminant represented by the formula (I), the value of the expression level in the specimen is assigned to x_i (i = 1, 2, ..., 42) of the formula (II) in sequence, to calculate solution D. When the solution D is a positive value, it can be determined that risk of late recurrence of the subject is high. When the solution D is zero or a negative value, it can be determined that the risk of late recurrence of the subject is low. For a subject whose risk of late recurrence has been determined to be high, it may be decided to administer hormones until 10 years after surgery. For a subject whose risk of late recurrence has been determined to be low, it may be decided to administer hormones until 5 years after surgery.
• The above 42 probe sets correspond to genes whose expression levels are significantly different between the late recurrence group and the early recurrence group. Therefore, when it has been determined that the risk of late recurrence is low based on the expression levels of genes measured by the probe sets, it means the same as it has been determined that the risk of early recurrence is high. Accordingly, in the present embodiment, a determination result that the risk of late recurrence is low may be rephrased as a determination result that the risk of early recurrence is high. That is, by performing the method of the present embodiment immediately after surgery, it can be predicted whether the risk of late recurrence is high or the risk of early recurrence is high for a subject. When breast cancer does not recur for 5 years after surgery in the subject whose risk of late recurrence has been determined to be low (i.e., the subject whose risk of early recurrence has been determined to be high), it may be considered that breast cancer has been cured.
• In the present embodiment, collection of a biological sample, measurement of gene expression levels, and analysis of the gene expression levels may be performed at substantially the same time. In another embodiment, collection of a biological sample and measurement and analysis of gene expression levels may be performed at times apart by a predetermined period. For example, cells or tissues collected from a subject are stored as an FFPE specimen, and after a predetermined period (for example, 5 years after surgery) has elapsed, measurement and analysis of expression levels of the genes may be performed using an RNA sample prepared from the FFPE specimen. Alternatively, collection of a biological sample, measurement of gene expression levels, and analysis of the gene expression levels may be performed at times apart by a predetermined period. For example, an RNA sample is prepared from cells or tissues collected from a subject, expression levels of the genes are measured, measurement data is stored, and after a predetermined period (for example, 5 years after surgery) has elapsed, analysis of the expression levels may be performed using the measurement data.
• Since the method of the present embodiment can accurately predict the risk of late recurrence of breast cancer, it is possible to predict the recurrence risk within 10 years after surgery, by combining the method of the present embodiment with a publicly known method for predicting the risk of early recurrence of breast cancer. For example, the method of the present embodiment and the method for predicting the risk of early recurrence of breast cancer may be performed at the same time. Alternatively, first, the method for predicting the risk of early recurrence of breast cancer is performed, and when it has been determined that the recurrence risk within 5 years after surgery is low, the method of the present embodiment may be performed. Alternatively, when it is determined that the recurrence risk within 5 years after surgery is low by the method for predicting the risk of early recurrence of breast cancer, and the subject does not actually have recurrence of breast cancer for 5 years after surgery, the method of the present embodiment may be performed at 5 years after surgery. In the present embodiment, when it has been determined that the recurrence risk within 5 years after surgery is high by the method for predicting the risk of early recurrence of breast cancer, the method of the present embodiment may be further performed.
• When the method of the present embodiment is combined with a publicly known method for predicting the risk of early recurrence of breast cancer, the expression levels of genes used in the method of the present embodiment and the expression levels of genes used in the method for predicting the risk of early recurrence may be acquired by a single measurement, for example, with a microarray. In this case, all the measured expression levels of genes may be simultaneously analyzed. Alternatively, the expression levels of genes used in the method for predicting the risk of early recurrence are analyzed, then the expression levels of genes used in the method of the present embodiment may be analyzed.
• When a part of cells or tissues collected from a subject is stored as an FFPE specimen, first, from an RNA sample prepared from the cells or tissues, measurement and analysis of expression levels of the genes used in the method for predicting the risk of early recurrence are performed. After that (for example, after 5 years), from the RNA sample prepared from the FFPE specimen, the expression levels of genes used in the method of the present embodiment may be measured and analyzed.
• A publicly known method for predicting the risk of early recurrence of breast cancer is not particularly limited, and for example, the method described in Patent Document 1 is preferable (Patent Document 1 is incorporated herein by reference). In the present embodiment, the subject may be a subject whose risk of early recurrence of breast cancer has been determined to be low, based on gene expression levels measured using probe sets specified by 95 probe set IDs shown in Table 3-1 and Table 3-2.
[Table 5]
• Table 3-1

  Number	Probe set ID	Gene symbol	UniGene ID	GenBank accession number
  1	219306_at	KIF15	Hs.646856	NM_020242
  2	218585_s_at	DTL	Hs.656473	NM_016448
  3	221677_s_at	DONSON	Hs.436341	AF232674
  4	201088_at	KPNA2	Hs.594238	NM_002266
  5	209034_at	PNRC1	Hs.75969	AF279899
  6	202610_s_at	MED14	Hs.407604	AF135802
  7	218906_x_at	KLC2	Hs.280792	NM_022822
  8	212723_at	JMJD6	Hs.514505	AK021780
  9	222231_s_at	LRRC59	Hs.370927	AK025328
  10	208838_at	CAND1	Hs.546407	AB020636
  11	218039_at	NUSAP1	Hs.615092	NM_016359
  12	209472_at	CCBL2	Hs.481898	BC000819
  13	212898_at	KIAA0406	Hs.655481	AB007866
  14	202620_s_at	PLOD2	Hs.477866	NM_000935
  15	201059_at	CTTN	Hs.596164	NM_005231
  16	201841_s_at	HSPB1	Hs.520973	NM_001540
  17	203755_at	BUB1B	Hs.631699	NM_001211
  18	211750_x_at	TUBA1C	Hs.719091	BC005946
  19	38158_at	ESPL1	Hs.153479	D79987
  20	204709_s_at	KIF23	Hs.270845	NM_004856
  21	201589_at	SMC1A	Hs.211602	D80000
  22	218460_at	HEATR2	Hs.535896	NM_017802
  23	207430_s_at	MSMB	Hs.255462	NM_002443
  24	212139_at	GCN1L1	Hs.298716	D86973
  25	211596_s_at	LRIG1	Hs.518055	AB050468
  26	212160_at	XPOT	Hs.85951	AI984005
  27	219238_at	PIGV	Hs.259605	NM_017837
  28	203432_at	TMPO	Hs.11355	AW272611
  29	201377_at	UBAP2L	Hs.490551	NM_014847
  30	218875_s_at	FBXO5	Hs.520506	NM_012177
  31	221922_at	GPSM2	Hs.584901	AW195581
  32	218727_at	SLC38A7	Hs.10499	NM_018231
  33	207469_s_at	PIR	Hs.495728	NM_003662
  34	218483_s_at	C11orf60	Hs.533738	NM_020153
  35	204641_at	NEK2	Hs.153704	NM_002497
  36	219502_at	NEIL3	Hs.405467	NM_ 018248
  37	209054_s_at	WHSC1	Hs.113876	AF083389
  38	220318_at	EPN3	Hs.670090	NM_017957
  39	210297_s_at	MSMB	Hs.255462	U22178
  40	209186_at	ATP2A2	Hs.506759	M23114
  41	219787_s_at	ECT2	Hs.518299	NM_018098
  42	45633_at	GINS3	Hs.47125	AI421812
  43	200848_at	AHCYL1	Hs.705418	AA479488
  44	200822_x_at	TPI1	Hs.524219	NM_000365
  45	211072_x_at	TUBA1B	Hs.719075	BC006481
  46	200811_at	CIRBP	Hs.634522	NM_001280
  47	202864_s_at	SP100	Hs.369056	NM_003113
  48	202154_x_at	TUBB3	Hs.511743	NM_006086
  49	213152_s_at	SFRS2B	Hs.476680	AI343248
  50	209368_at	EPHX2	Hs.212088	AF233336

[Table 6]
• Table 3-2

  Number	Probe set ID	Gene symbol	UniGene ID	GenBank accession number
  51	211058_x_at	TUBA1B	Hs.719075	BC006379
  52	209251_x_at	TUBA1C	Hs.719091	BC004949
  53	213646_x_at	TUBA1B	Hs.719075	BE300252
  54	204540_at	EEF1A2	Hs.433839	NM_001958
  55	202026_at	SDHD	Hs.719164	NM_003002
  56	201090_x_at	TUBA1B	Hs.719075	NM_006082
  57	213119_at	SLC36A1	Hs.269004	AW058600
  58	217840_at	DDX41	Hs.484288	NM_016222
  59	206559_x_at	EEF1A1	---	NM_001403
  60	202066_at	PPFIA1	Hs.530749	AA195259
  61	203108_at	GPRC5A	Hs.631733	NM_003979
  62	218697_at	NCKIPSD	Hs.655006	NM_016453
  63	222039_at	KIF18B	Hs.135094	AA292789
  64	202069_s_at	IDH3A	Hs.591110	AI826060
  65	203362_s_at	MAD2L1	Hs.591697	NM_002358
  66	202666_s_at	ACTL6A	Hs.435326	NM_004301
  67	204892_x_at	EEF1A1	Hs.520703	NM_001402
  68	205682_x_at	APOM	Hs.534468	NM_019101
  69	209714_s_at	CDKN3	Hs.84113	AF213033
  70	218381_s_at	U2AF2	Hs.528007	NM_007279
  71	201947_s_at	CCT2	Hs.189772	NM_006431
  72	212722_s_at	JMJD6	Hs.514505	AK021780
  73	204825_at	MELK	Hs.184339	NM_014791
  74	203184_at	FBN2	Hs.519294	NM_001999
  75	201266_at	TXNRD1	Hs.708065	NM_003330
  76	202969_at	DYRK2	Hs.173135	AI216690
  77	204817_at	ESPL1	Hs.153479	NM_012291
  78	209523_at	TAF2	Hs.122752	AK001618
  79	218491_s_at	THYN1	Hs.13645	NM_014174
  80	217363_x_at	---	---	AL031313
  81	218009_s_at	PRC1	Hs.567385	NM_003981
  82	204026_s_at	ZWINT	Hs.591363	NM_007057
  83	218355_at	KIF4A	Hs.648326	NM_012310
  84	202153_s_at	NUP62	Hs.574492	NM_016553
  85	213011_s_at	TPI1	Hs.524219	BF116254
  86	217966_s_at	FAM129A	Hs.518662	NM_022083
  87	214782_at	CTTN	Hs.596164	AU155105
  88	217967_s_at	FAM129A	Hs.518662	AF288391
  89	204649_at	TROAP	Hs.524399	NM_005480
  90	35671_at	GTF3C1	Hs.371718	U02619
  91	213502_x_at	LOC91316	Hs.148656	AA398569
  92	221285_at	ST8SIA2	Hs.302341	NM_006011
  93	221519_at	FBXW4	Hs.500822	AF281859
  94	20255_1_s_at	CRIM1	Hs.699247	BG546884
  95	217138_x_at	IGL@	Hs.449585	AJ249377

• When the subject is administered with anticancer agents as a preoperative adjuvant therapy, the method of the present embodiment may be combined with a publicly known method for predicting responsiveness to chemotherapy and prognosis. As such a method, a multigene assay, MPCP155 (see Tsunashima R. et al., Construction of multi-gene classifier for prediction of response to and prognosis after neoadjuvant chemotherapy for estrogen receptor positive breast cancers, Cancer Letters, 2015, vol. 365, p. 166-173, which is incorporated herein by reference) is known. In the present embodiment, the subject may be a subject whose responsiveness to chemotherapy has been determined, based on gene expression levels measured using probe sets specified by 155 probe set IDs shown in Table 4-1 and Table 4-5.
[Table 7]
• Table 4-1

  Number	Probe set ID	Gene symbol
  1	205896_at	SLC22A4
  2	203685_at	BCL2
  3	200970_s_at	SERP1
  4	205355_at	ACADSB
  5	210678_s_at	AGPAT2
  6	210046_s_at	IDH2
  7	37152_at	PPARD
  8	205733_at	BLM
  9	215649_s_at	MVK
  10	214594_x_at	ATP8B1
  11	218096_at	AGPAT5
  12	203684_s_at	BCL2
  13	202671_s_at	PDXK
  14	32837_at	AGPAT2
  15	218923_at	CTBS
  16	200846_s_at	PPP1CA
  17	207005_s_at	BCL2
  18	221753_at	SSH1
  19	208705_s_at	EIF5
  20	207621_s_at	PEMT
  21	200849_s_at	AHCYL1
  22	221065_s_at	CHST8
  23	209281_s_at	ATP2B1
  24	217006_x_at	FASN
  25	214095_at	SHMT2
  26	213607_x_at	NADK
  27	206130_s_at	ASGR2
  28	200784_s_at	LRP1
  29	205301_s_at	OGG1
  30	201627_s_at	INSIG1
  31	205768_s_at	SLC27A2
  32	203770_s_at	STS
  33	202376_at	SERPINA3
  34	219723_x_at	AGPAT3
  35	218506_x_at	GLYR1
  36	208051_s_at	PAIP1
  37	205769_at	SLC27A2

[Table 8]
• Table 4-2

  Number	Probe set ID	Gene symbol
  38	211762_s_at	KPNA2
  39	204879_at	PDPN
  40	219429_at	FA2H
  41	211627_x_at	ESR1
  42	202779_s_at	UBE2S
  43	210609_s_at	TP53I3
  44	209064_x_at	PAIP1
  45	208830_s_at	SUPT6H
  46	219048_at	PIGN
  47	209041_s_at	UBE2G2
  48	201523_x_at	UBE2N
  49	208511_at	PTTG3P
  50	222074_at	UROD
  51	209496_at	RARRES2
  52	203366_at	POLG
  53	201782_s_at	AIP
  54	202715_at	CAD
  55	208624_s_at	EIF4G1
  56	210449_x_at	MAPK14
  57	201610_at	ICMT
  58	217294_s_at	ENOI
  59	205140_at	FPGT
  60	204430_s_at	SLC2A5
  61	204167_at	BTD
  62	203566_s_at	AGL
  63	210688_s_at	CPT1A
  64	213613_s_at	NADK
  65	215988_s_at	DLG1
  66	206925_at	ST8SIA4
  67	203365_s_at	MMP15
  68	215842_s_at	ATP11A
  69	203939_at	NT5E
  70	201695_s_at	PNP
  71	200878_at	EPAS1
  72	209740_s_at	PNPLA4
  73	218313_s_at	GALNT7
  74	209773_s_at	RRM2

[Table 9]
• Table 4-3

  Number	Probe set ID	Gene symbol
  75	217607_x_at	EIF4G2
  76	205816_at	ITGB8
  77	203244_at	PEX5
  78	207851_s_at	INSR
  79	207275_s_at	ACSL1
  80	204044_at	QPRT
  81	201218_at	CTBP2
  82	209916_at	DHTKD1
  83	200785_s_at	LRP1
  84	213343_s_at	GDPD5
  85	218533_s_at	UCKL1
  86	203381_s_at	APOE
  87	221142_s_at	PECR
  88	209533_s_at	PLAA
  89	218018_at	PDXK
  90	205225_at	ESR1
  91	32502_at	GDPD5
  92	201080_at	PIP4K2B
  93	208696_at	CCT5
  94	209735_at	ABCG2
  95	203554_x_at	PTTG1
  96	206587_at	CCT6B
  97	202580_x_at	FOXM1
  98	214210_at	SLC25A17
  99	201291_s_at	TOP2A
  100	200675_at	CD81
  101	208510_s_at	PPARG
  102	205128_x_at	PTGS1
  103	216551_x_at	PLCG1
  104	203634_s_at	CPT1A
  105	204946_s_at	TOP3A
  106	212883_at	APOE
  107	207076_s_at	ASS1
  108	211026_s_at	MGLL
  109	200879_s_at	EPAS1
  110	209537_at	EXTL2
  111	221485_at	B4GALT5

[Table 10]
• Table 4-4

  Number	Probe set ID	Gene symbol
  112	220948_s_at	ATP1A1
  113	218760_at	COQ6
  114	211233_x_at	ESR1
  115	203343_at	UGDH
  116	217190_x_at	ESR1
  117	203099 s at	CDYL
  118	202533_s_at	DHFR
  119	206197_at	NME5
  120	219718_at	FGGY
  121	201081_s_at	PIP4K2B
  122	215210_s_at	DLST
  123	211234_x_at	ESR1
  124	203767_s_at	STS
  125	218924_s_at	CTBS
  126	204059_s_at	ME1
  127	211754_s_at	SLC25A17
  128	213279_at	DHRS1
  129	217289_s_at	SLC37A4
  130	204161_s_at	ENPP4
  131	34187_at	RBMS2
  132	202922_at	GCLC
  133	209616_s_at	CES1
  134	218951_s_at	PLCXD1
  135	40829_at	WDTC1
  136	201117_s_at	CPE
  137	213379_at	COQ2
  138	203633_at	CPT1A
  139	206311_s_at	PLA2G1B
  140	206630_at	TYR
  141	210745_at	ONECUT1
  142	205552_s_at	OAS1
  143	209425_at	AMACR
  144	201625_s_at	INSIG1
  145	203072_at	MYOIE
  146	209998_at	PIGO
  147	201898_s_at	UBE2A
  148	204956_at	MTAP

[Table 11]
• Table 4-5

  Number	Probe set ID	Gene symbol
  149	215966_x_at	GK3P
  150	209681_at	SLC19A2
  151	59631_at	TXNRD3
  152	219394_at	PGS1
  153	206492_at	FHIT
  154	211738_x_at	CELA3A
  155	209165_at	AATF

• Flows when a treatment policy for breast cancer patients is decided by combining the method of the present embodiment with a publicly known multigene assay for predicting risk of early recurrence of breast cancer will be described with reference to Figs. 1 to 4. In any of Figs. 1 to 4, first, a biological sample is collected from a subject by vacuum-assisted breast biopsy (VAB). Then, tissue diagnosis is performed using a part of the biological sample to diagnose breast cancer. When diagnosed as breast cancer, RNA samples are prepared from the remaining biological sample, and expression levels of genes are measured using GeneChip (registered trademark) Human Genome U133 Plus 2.0 Array (Affymetrix, Inc.). This microarray contains not only the 42 probe sets used in the method of the present embodiment but also probe sets used in Curebest (registered trademark) 95GC Breast (hereinafter also referred to as "95GC") and MPCP155 (hereinafter also referred to as "155GC"). Accordingly, it is possible to acquire gene expression level data for a plurality of multigene assays by performing measurement by microarray only once, with respect to a biological sample obtained by biopsy for diagnosis of breast cancer. A multigene assay by microarray may further be combined with a multigene assay by RT-PCR method. The multigene assay by RT-PCR method includes Oncotype DX (registered trademark). In this case, first, tissue diagnosis is performed using a part of the biological sample obtained by VAB to diagnose breast cancer. When diagnosed as breast cancer, RNA samples are prepared from the remaining biological sample, and expression levels of genes are measured using GeneChip (registered trademark) Human Genome U133 Plus 2.0 Array (Affymetrix, Inc.) and RT-PCR method.
• Referring to Fig. 1, first, a biological sample is collected from a subject by VAB, and diagnosis of breast cancer by tissue diagnosis and determination of risk of early recurrence by 95GC are performed. When it has been determined that the risk of early recurrence is low by 95GC, the process proceeds to step S1-1. In step S1-1, preoperative hormone therapy (NAE) or surgery is performed on the subject. When it has been determined that the risk of early recurrence is high by 95GC, the process proceeds to step S1-2. In step S1-2, preoperative chemotherapy (NAC (A/T)) with adriamycin or a taxane anticancer agent is performed on the subject, and then surgery is performed. When the postoperative course is pathologically complete remission (pCR), the process proceeds to step S1-3. In step S1-3, determination of risk of late recurrence by the method of the present embodiment (hereinafter also referred to as "42GC") is performed. When it has been determined that the risk of late recurrence is low by 42GC, the process proceeds to step S1-4. In step S1-4, hormone therapy for 5 years after surgery (5y-HT) is performed on the subject. When it has been determined that the risk of late recurrence is high by 42GC, the process proceeds to step S1-5. In step S1-5, hormone therapy for 10 years after surgery (10y-HT) is performed on the subject.
• When the postoperative course in step S1-2 is pathologically complete non-remission (Non-pCR), the process proceeds to step S1-6. In step S1-6, determination of responsiveness to chemotherapy by 155GC is performed. When it has been determined that the responsiveness is low (L-CS*: low chemo-sensitivity) by 155GC, the process proceeds to step S1-7. In step S1-7, determination of the risk of late recurrence by 42GC is performed. When it has been determined that the risk of late recurrence is low by 42GC, the process proceeds to step S1-8, and when it has been determined that the risk of late recurrence is high, the process proceeds to step S1-9. Steps S1-8 and S1-9 are the same as those described for steps S1-4 and S1-5, respectively. In step S1-6, when it has been determined that the responsiveness is high (H-CS*: high chemo-sensitivity) by 155GC, the process proceeds to step S1-10. In step S1-10, additional chemotherapy (Additional CT) is performed on the subject and determination of the risk of late recurrence by 42GC is performed. When it has been determined that the risk of late recurrence is low by 42GC, the process proceeds to step S1-11, and when it has been determined that the risk of late recurrence is high, the process proceeds to step S1-12. Steps S1-11 and S1-12 are the same as those described for steps S1-4 and S1-5, respectively.
• Referring to Fig. 2, first, a biological sample is collected from a subject by VAB, and diagnosis of breast cancer is performed by tissue diagnosis. When diagnosed as breast cancer, the process proceeds to step S2-1. In step S2-1, surgery is performed on the subject. When it has been determined that there is no lymph node metastasis (pN0) by postoperative pathological classification (pN classification), determination of the risk of early recurrence by 95GC is performed. When it has been determined that the risk of early recurrence is low by 95GC, the process proceeds to step S2-2. In step S2-2, determination of the risk of late recurrence by 42GC is performed on the subject. When it has been determined that the risk of late recurrence is low by 42GC, the process proceeds to step S2-3, and when it has been determined that the risk of late recurrence is high, the process proceeds to step S2-4. Steps S2-3 and S2-4 are the same as those described for steps S1-4 and S1-5, respectively. In step S2-1, when it has been determined that the risk of early recurrence is high by 95GC, the process proceeds to step S2-5. In step S2-5, chemotherapy (CT) is performed on the subject and determination of the risk of late recurrence by 42GC is performed. When it has been determined that the risk of late recurrence is low by 42GC, the process proceeds to step S2-6, and when it has been determined that the risk of late recurrence is high, the process proceeds to step S2-7. Steps S2-6 and S2-7 are the same as those described for steps S1-4 and S1-5, respectively.
• Referring to Fig. 3, first, a biological sample is collected from a subject by VAB, and diagnosis of breast cancer is performed by tissue diagnosis. When diagnosed as breast cancer, the process proceeds to step S3-1. In step S3-1, surgery is performed on the subject. When it has been determined that there is lymph node metastasis (pN1-3) by pN classification, the process proceeds to step S3-2. In step S3-2, chemotherapy (CT) is performed on the subject and determination of responsiveness to chemotherapy by 155GC is performed. When it has been determined that the responsiveness is low (L-CS*) by 155GC, the process proceeds to step S3-3. In step S3-3, determination of the risk of late recurrence by 42GC is performed. When it has been determined that the risk of late recurrence is low by 42GC, the process proceeds to step S3-4, and when it has been determined that the risk of late recurrence is high, the process proceeds to step S3-5. Steps S3-4 and S3-5 are the same as those described for steps S1-4 and S1-5, respectively In step S3-2, when it has been determined that the responsiveness is high (H-CS*) by 155GC, the process proceeds to step S3-6. In step S3-6, additional chemotherapy (Additional CT) is performed on the subject and determination of the risk of late recurrence by 42GC is performed. When it has been determined that the risk of late recurrence is low by 42GC, the process proceeds to step S3-7, and when it has been determined that the risk of late recurrence is high, the process proceeds to step S3-8. Steps S3-7 and S3-8 are the same as those described for steps S1-4 and S1-5, respectively
• Referring to Fig. 4, first, a biological sample is collected from a subject by VAB, and diagnosis of breast cancer is performed by tissue diagnosis. When diagnosed as breast cancer, the process proceeds to step S4-1. In step S4-1, surgery is performed on the subject. When it has been determined that there is lymph node metastasis (pN1) by pN classification, determination of the risk of early recurrence by Oncotype DX (registered trademark) is performed. When it has been determined that the risk of early recurrence is low by Oncotype DX (registered trademark), the process proceeds to step S4-2. In step S4-2, determination of the risk of late recurrence by 42GC is performed on the subject. When it has been determined that the risk of late recurrence is low by 42GC, the process proceeds to step S4-3, and when it has been determined that the risk of late recurrence is high, the process proceeds to step S4-4. Steps S4-3 and S4-4 are the same as those described for steps S1-4 and S1-5, respectively In step S4-1, when it has been determined that the risk of early recurrence is high by Oncotype DX (registered trademark), the process proceeds to step S4-5. In step S4-5, chemotherapy (CT) is performed on the subject and determination of responsiveness to chemotherapy by 155GC is performed. When it has been determined that the responsiveness is low (L-CS*) by 155GC, the process proceeds to step S4-6. In step S4-6, determination of the risk of late recurrence by 42GC is performed. When it has been determined that the risk of late recurrence is low by 42GC, the process proceeds to step S4-7, and when it has been determined that the risk of late recurrence is high, the process proceeds to step S4-8. Steps S4-7 and S4-8 are the same as those described for steps S1-4 and S1-5, respectively In step S4-5, when it has been determined that the responsiveness is high (H-CS*) by 155GC, the process proceeds to step S4-9. In step S4-9, additional chemotherapy (Additional CT) is performed on the subject and determination of the risk of late recurrence by 42GC is performed. When it has been determined that the risk of late recurrence is low by 42GC, the process proceeds to step S4-10, and when it has been determined that the risk of late recurrence is high, the process proceeds to step S4-11. Steps S4-10 and S4-11 are the same as those described for steps S1-4 and S1-5, respectively
[2. Determination Device and Computer Program]
• The scope of the present disclosure also includes a device and a computer program for implementing the above-described method of the present embodiment. An example of the device for determining prognosis of breast cancer of the present embodiment will be described with reference to the drawings. However, the present embodiment is not limited only to the embodiment shown in this example. A determination device 10 shown in Fig. 5 includes a measuring device 20 and a computer system 30 connected to the measuring device 20.
• In this example, the measuring device 20 is a microarray scanner that detects a signal based on a target nucleic acid hybridized to a probe on a microarray. The signal is optical information such as a fluorescence signal. In this case, when the microarray that is brought into contact with the RNA sample is set in the measuring device 20, the measuring device 20 acquires optical information based on the target nucleic acid hybridized to the probe on the microarray, and the measuring device 20 transmits the acquired optical information to the computer system 30.
• The microarray scanner is not particularly limited as long as it can detect a signal based on the target nucleic acid hybridized to the probe. Since the type of the signal differs depending on the labeling substance used for labeling the target nucleic acid, the microarray scanner can be appropriately selected according to the type of the labeling substance. For example, when the labeling substance is a fluorescent substance, a microarray scanner capable of detecting a fluorescence signal can be used as the measuring device 20.
• When expression levels of genes are analyzed by a nucleic acid amplification method such as quantitative RT-PCR, the measuring device 20 may be a nucleic acid amplification detection apparatus. In this case, a reaction solution containing an RNA sample, an enzyme for nucleic acid amplification, a primer and the like is set in the measuring device 20, and a nucleic acid in the reaction solution is amplified by the nucleic acid amplification method. The measuring device 20 acquires optical information such as fluorescence generated from the reaction solution and turbidity of the reaction solution by an amplification reaction, and the measuring device 20 transmits the optical information to the computer system 30.
• The computer system 30 includes a computer main body 300, an input unit 301, and a display unit 302 that displays specimen information, a determination result, and the like. The computer system 30 receives the optical information from the measuring device 20. Then, the processor of the computer system 30 executes a program for determining prognosis of breast cancer, based on the optical information. As shown in Fig. 5, the computer system 30 may be equipment separate from the measuring device 20, or may be equipment including the measuring device 20. In the latter case, the computer system 30 may itself be the determination device 10.
• Referring to Fig. 6, the computer main body 300 includes a central processing unit (CPU) 310, a read only memory (ROM) 311, a random access memory (RAM) 312, a hard disk 313, an input/output interface 314, a reading device 315, a communication interface 316, and an image output interface 317. The CPU 310, the ROM 311, the RAM 312, the hard disk 313, the input/output interface 314, the reading device 315, the communication interface 316 and the image output interface 317 are data-communicably connected by a bus 318. The measuring device 20 is communicably connected to the computer system 30 via the communication interface 316.
• The CPU 310 can execute a program stored in the ROM 311 or the hard disk 313 and a program loaded in the RAM 312. The CPU 310 calculates fluorescence intensity based on the optical information acquired from the measuring device 20. The CPU 310 calculates solution D according to a discriminant represented by formula (I) stored in the ROM 311 or the hard disk 313. Then, the CPU 310 determines prognosis of breast cancer based on the acquired solution D and the determination criteria stored in the ROM 311 or the hard disk 313. The CPU 310 outputs the determination result and displays the determination result on the display unit 302.
• The ROM 311 includes a mask ROM, PROM, EPROM, EEPROM, and the like. In the ROM 311, a computer program executed by the CPU 310 and data used for executing the computer program are recorded.
• The RAM 312 includes SRAM, DRAM, and the like. The RAM 312 is used for reading the program recorded in the ROM 311 and the hard disk 313. The RAM 312 is also used as a work area of the CPU 310 when these programs are executed.
• The hard disk 313 has installed therein an operating system to be executed by the CPU 310, a computer program such as an application program (the program for determining prognosis of breast cancer), and data used for executing the computer program.
• The reading device 315 includes a flexible disk drive, a CD-ROM drive, a DVD-ROM drive, and the like. The reading device 315 can read a program or data recorded on a portable recording medium 40.
• The input/output interface 314 includes, for example, a serial interface such as USB, IEEE1394 and RS-232C, a parallel interface such as SCSI, IDE and IEEE1284, and an analog interface including a D/A converter, an A/D converter and the like. The input unit 301 such as a keyboard and a mouse is connected to the input/output interface 314. An operator can input various commands to the computer main body 300 through the input unit 301.
• The communication interface 316 is, for example, an Ethernet (registered trademark) interface or the like. The computer main body 300 can also transmit print data to a printer or the like through the communication interface 316.
• The image output interface 317 is connected to the display unit 302 including an LCD, a CRT, and the like. As a result, the display unit 302 can output a video signal corresponding to the image data coming from the CPU 310. The display unit 302 displays an image (screen) according to the input video signal.
• Referring to Fig. 7, a processing procedure for determining prognosis of breast cancer executed by the determination device 10 will be described. Here, a case where determination of the risk of late recurrence is performed based on a fluorescence signal generated from the target nucleic acid bound to the probe on the microarray will be described as an example. However, the present embodiment is not limited to this example.
• In step S101, the CPU 310 acquires optical information (fluorescence signal) from the measuring device 20, the CPU 310 calculates a fluorescence intensity from the acquired optical information, and the CPU 310 stores the calculated fluorescence intensity in the hard disk 313. In step S102, the CPU 310 calculates solution D according to the discriminant represented by the formula (I) stored in the hard disk 313, using the calculated fluorescence intensity, and the CPU 310 stores the solution D in the hard disk 313. In step S103, the CPU 310 compares the calculated solution D with the determination criteria stored in the hard disk 313. When the solution D is a positive value, the process proceeds to step S104, and a determination result indicating that the late recurrence risk of the subject is high is stored in the hard disk 313. When the solution D is zero or a negative value, the process proceeds to step S105, and a determination result indicating that the late recurrence risk of the subject is low is stored in the hard disk 313. In step S105, the CPU 310 outputs the determination result, and the CPU 310 displays the determination result on the display unit 302, or the CPU 310 makes a printer print out the determination result. Accordingly, it is possible to provide doctors and the like with information to assist the determination of prognosis of breast cancer.
• Hereinafter, the present disclosure will be described in more detail by examples, but the present disclosure is not limited to these examples.
EXAMPLES Example 1: Search for Predictors of Late Recurrence (1) Acquisition of gene expression levels
• Data of breast cancer patients (779 cases) shown in Table 5 were extracted from four data sets of accession numbers: GSE6532, GSE12093, GSE17705 and GSE26971 in NCBI Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) of microarray experiments. All of these patients were ER positive and also received only administration of hormones as a treatment. In the table, "cT" is a clinical tumor diameter, "cN" is a clinical nodal state (presence or absence of lymph node metastasis), and "PR" is a progesterone receptor. The term "95GC" indicates risk of early recurrence determined by Curebest (registered trademark) 95GC Breast (Sysmex Corporation). The term "155GC" indicates responsiveness to chemotherapy (chemo-sensitivity) determined by MPCP155 (see Non-Patent Document 1). Hereinbelow, among patients with recurrence (177 cases), patients with recurrence within 5 years (≤ 5 years) after surgery are called "early recurrence group", and patients with recurrence more than 5 years (> 5 years) after surgery are called "late recurrence group".
[Table 12]
• Table 5

  	Training set
  	GSE6532	GSE12093	GSE17705	GSE26971	Total
  Number of cases	87	136	298	258	779
  Age
  < 50	3	NA	NA	NA	3
  ≥ 50	84	NA	NA	NA	84
  cT
  T1	43	NA	NA	102	145
  T2/3/4	44	NA	NA	145	189
  Unknown	0	NA	NA	11	11
  cN
  Positive	58	0	112	87	257
  Negative	29	136	175	129	469
  Unknown	0	0	11	42	53
  Histological grade
  1	17	NA	NA	NA	17
  2/3	53	NA	NA	NA		53
  Unknown	17	NA	NA	NA	17
  PR
  Positive	64	NA	NA	NA	64
  Negative	21	NA	NA	NA	21
  Unknown	2	NA	NA	NA		2
  Recurrence	28	20	71	58	177
  ≤ 5 yrs	14	12	42	41	109
  > 5 yrs	14	8	29	17	68
  No recurrence Sub type	59	116	227	200	602
  Luminal A	37	65	148	132	382
  Luminal B	50	71	150	126	397
  95GC
  Low risk
  	48	85	177	151	461
  High risk	39	51	121	107	318
  155GC
  Low sensitivity
  	42	61	146	113	362
  High sensitivity	45	75	152	145	417

(2) Analysis of gene expression level data
• Gene expression level data (fluorescence intensity data) of the breast cancer patients was normalized by using CEL file data of each data set and MAS5 statistical algorithm of analysis software (Affymetrix Expression Console (trademark) software, manufactured by Affymetrix, Inc.). Next, for each data set, from a value of a gene expression level measured by each probe set of the microarray, a mean value of the gene expression levels in the data set was subtracted to standardize the value of the expression level of each gene (mean-centering). Then, z-score was calculated for each probe set on the microarray, by using a package "GeneMeta v1 .16.0" (http://www.bioconductor.org/packages/2.4/bioc/html/GeneMeta.html) contained in an additional package "BioConductor" ver. 2.4 used in software for statistical analysis "R", according to a literature by Choi J.K et al. (Combining multiple microarray studies and modeling interstudy variation, Bioinformatics, 2003, vol. 19, suppl. 1, p. i84-94).
• It was evaluated by chi-square test that the expression level of the gene corresponding to the probe set was different between the early recurrence group and the late recurrence group, and a probe set with p value of less than 0.01 was selected. Next, using a diagonal linear discriminant analysis (DLDA) as an algorithm of a discriminant, a probe set group was selected while increasing the number of the probe sets to be selected one by one in ascending order of the p value in the chi-square test, and a discriminant was constructed. Using the obtained discriminant and data of the above 177 recurrent patients, the number of probe sets with which accuracy and negative predictive value are maximized were determined by Leave-One-Out Cross-Validation method. The accuracy refers to a ratio obtained by dividing the sum of "the number of cases in which late recurrence was predicted, and late recurrence occurred" and "the number of cases in which late recurrence was not predicted, and late recurrence did not occur" by "the total number of cases". The negative predictive value is a ratio obtained by dividing "the number of cases in which late recurrence was not predicted, and late recurrence did not occur" by "the number of cases in which late recurrence was not predicted". The results are shown in Fig. 8.
(3) Results
• From Fig. 8, it was found that the negative predictive value and the accuracy were maximized when the number of probe sets was 42. It was found that the negative predictive value was at least 70% when the number of probe sets was nine or more. As described above, the negative predictive value is a probability that late recurrence actually does not occur when it is predicted that late recurrence does not occur. The selected 42 probe sets correspond to genes whose expression levels were significantly different between the late recurrence group and the early recurrence group. Therefore, when it is predicted that late recurrence does not occur based on the expression levels of genes measured by the probe sets, it means the same as the case where it is predicted that early recurrence occurs. That is, the negative predictive value can be rephrased as a probability that early recurrence actually occurs when it is predicted that early recurrence occurs.
• IDs, z-scores, p values of chi-square test and weighting factors in the discriminants of the selected 42 probe sets are shown in Table 6-1 and Table 6-2. In which of the late recurrence group and the early recurrence group each gene is highly expressed is also shown in Table 6-1 and Table 6-2. In the table, "late" refers to a late recurrence group, "early" refers to an early recurrence group. The probe set IDs are IDs assigned to the probe sets of GeneChip (registered trademark) Human Genome U133 Plus 2.0 Array (Affymetrix, Inc.).
[Table 13]
• Table 6-1

  Gene number	Probe set ID	Gene symbol	Z-score	P value	Weighting factor	High expression
  1	212196_at	IL6ST	-4.12792	0.0000366	1.43235997	Late
  2	205440_s_at	NPY1R	-3.88231	0.000103	0.450736337	Late
  3	208788_at	ELOVL5	-3.72932	0.000192	1.793041948	Late
  4	214306_at	OPA1	3.70520	0.000211	-2.545950461	Early
  5	212195_at	IL6ST	-3.66917	0.000243	1.539087017	Late
  6	210980_s_at	ASAH1	-3.65969	0.000252	1.539921375	Late
  7	204864_s_at	IL6ST	-3.65831	0.000253	1.151743979	Late
  8	221590_s_at	ALDH6A1	-3.59018	0.000330	1.837908119	Late
  9	218692_at	SYBU	-3.57397	0.000351	1.149172541	Late
  10	205428_s_at	CALB2	3.55933	0.000371	-1.179415269	Early
  11	206414_s_at	ASAP2	3.55143	0.000383	-2.323532179	Early
  12	201156_s_at	RAB5C	-3.54402	0.000394	2.316847761	Late
  13	216988_s_at	PTP4A2	-3.52418	0.000424	2.002072892	Late
  14	213485_s_at	ABCC10	3.51999	0.000431	-2.306701292	Early
  15	206825_at	OXTR	3.51415	0.000441	-1.1298499	Early
  16	211538_s_at	HSPA2	-3.50756	0.000452	0.964294299	Late
  17	205596_s_at	SMURF2	3.49837	0.000468	-2.109236485	Early
  18	202752_x_at	SLC7A8	-3.46945	0.000521	1.363057605	Late
  19	214435_x_at	RALA	3.45260	0.000555	-2.288374092	Early
  20	209869_at	ADRA2A	-3.44181	0.000577	1.377444437	Late
  21	203360_s_at	MYCBP	-3.39767	0.000679	1.688897659	Late
  22	205898_at	CX3CR1	-3.39344	0.000690	0.740592577	Late
  23	203720_s_at	ERCC1	-3.37674	0.000733	1.658860215	Late
  24	205963_s_at	DNAJA3	-3.36349	0.000769	2.534304027	Late
  25	211000_s_at	IL6ST	-3.31557	0.000914	0.971312607	Late
  26	203045_at	NINJ1	-3.29288	0.000991	1.889380552	Late
  27	218513_at	C4orf43	-3.26230	0.001105	1.429869188	Late
  28	209417_s_at	IFI35	-3.26153	0.001108	1.323441517	Late
  29	213527_s_at	ZNF688	-3.25889	0.001118	2.158913805	Late
  30	201716_at	SNX1	-3.25195	0.001146	2.2492475	Late
  31	201988_s_at	CREBL2	-3.22177	0.001274	1.921069642	Late
  32	204934_s_at	HPN	-3.21415	0.001308	1.115597062	Late
  33	204862_s_at	NME3	-3.13797	0.001701	1.661682832	Late
  34	203769_s_at	STS	3.12877	0.001755	-1.392023997	Early
  35	204334_at	KLF7	3.10363	0.001911	-1.662199312	Early
  36	208911_s_at	PDHB	-3.09973	0.001936	2.071027409	Late
  37	204863_s_at	IL6ST	-3.09878	0.001943	0.825924727	Late

[Table 14]
• Table 6-2

  Gene number	Probe set ID	Gene symbol	Z-score	P value	Weighting factor	High expression
  38	209706_at	NKX3-1	-3.08717	0.002020	0.611126615	Late
  39	203733_at	DEXI	-3.07032	0.002138	1.847912036	Late
  40	213702_x_at	ASAH1	-3.06744	0.002158	1.575733773	Late
  41	202554_s_at	GSTM3	-3.05803	0.002227	0.648154701	Late
  42	221515_s_at	LCMT1	-3.04479	0.002328	1.7259299	Late

• Based on the above results, a final discriminant was constructed. The obtained discriminant is represented by the following formula (I). $$\mathrm{D}={\mathrm{\Sigma }}_{\mathrm{i}}\left({\mathrm{w}}_{\mathrm{i}}\times {\mathrm{y}}_{\mathrm{i}}\right)-\left(-\mathrm{4.763756453}\right)$$

  

  wherein i represents a number assigned to each gene shown in Table 6-1 and Table 6-2, w_i represents a weighting factor of gene numbered i shown in Table 6-1 and Table 6-2, y_i represents a standardized expression level of gene according to a formula represented by formula (II): $${\mathrm{y}}_{\mathrm{i}}={\mathrm{x}}_{\mathrm{i}}-{\mathrm{m}}_{\mathrm{i}}$$

  

  wherein x_i represents an expression level of gene numbered i shown in Table 6-1 and Table 6-2, and m_i represents a mean value of expression levels of genes numbered i shown in Table 6-1 and Table 6-2 over specimens,
  and ∑_i represents a sum total over the genes.
• When solution D of the discriminant is a positive value, it is predicted that risk of late recurrence of the subject is high. When the solution D is zero or a negative value, it is predicted that the risk of late recurrence of the subject is low. When it is predicted that the risk of late recurrence is high, there is a possibility that micrometastasis of breast cancer with slow growth rate may exist in the body of the subject even if no recurrence is observed for 5 years after surgery, and extended endocrine treatment can be applied to the subject. As described above, the prediction that the risk of late recurrence is low can be rephrased as a prediction that the risk of early recurrence is high. When breast cancer does not recur for 5 years after surgery in a subject predicted to have a low risk of late recurrence, it can be considered that breast cancer has been cured.
Example 2: Verification of predictors of late recurrence
• Among a training set of Example 1, the risk of late recurrence of breast cancer was predicted for ER-positive breast cancer patients (564 cases) who did not recur for 5 years after surgery, by the discriminant represented by the formula (I). For patients (381 cases) predicted to have a high risk of late recurrence and patients (183 cases) predicted to have a low risk of late recurrence, the distant recurrence-free survival rate (DRFS rate) for 15 years after surgery was examined by Kaplan-Meier plot. Significant differences were evaluated by log-rank test. The results are shown in Fig. 9A.
• Gene expression level data of ER-positive breast cancer patients (165 cases) who did not recur for 5 years after surgery, which are different from the breast cancer patients of the training set, was used as a validation set. Using the data, the risk of late recurrence was predicted in the same manner as above. For patients (112 cases) predicted to have a high risk of late recurrence and patients (53 cases) predicted to have a low risk of late recurrence, the distant recurrence-free survival rate for 15 years after surgery was examined by Kaplan-Meier plot. Significant differences were evaluated by log-rank test. The results are shown in Fig. 9B.
• From Fig. 9A, the distant recurrence-free survival rate for 15 years after surgery was about 70% for patients predicted to have a high risk of late recurrence ("high-risk group" in the figure), whereas it was about 85% for patients predicted to have a low risk of late recurrence ("low-risk group" in the figure). p was 0.0061 by log-rank test. Likewise, from Fig. 9B, the distant recurrence-free survival rate for 15 years after surgery was about 75% in the high-risk group, whereas it was 100% in the low-risk group. p was 0.020 by log-rank test. Accordingly, it can be said that the risk of late recurrence is significantly lower in the low-risk group than in the high-risk group. From these results, it was shown that the risk of late recurrence of breast cancer can be predicted with high accuracy by analyzing the gene expression levels measured using the probes specified by the probe set IDs shown in Table 6-1 and Table 6-2.

Claims (15)

1. A method for acquiring information on prognosis of breast cancer, the method comprising:

   a measuring step of measuring expression levels of genes of IL6ST, NPY1R, ELOVL5, OPA1, ASAH1, ALDH6A1 and SYBU in an RNA sample prepared from a biological sample collected from a subject;

   an analyzing step of analyzing the measured expression levels of genes; and

   an acquiring step of acquiring information on prognosis of breast cancer based on an analysis result.
2. The method according to claim 1, wherein, in the measuring step, the expression levels are measured using probe sets specified by probe set IDs shown in Table 1.
   [Table 1] Table 1 Gene number Probe set ID Gene symbol 1 212196_at IL6ST 2 205440_s_at NPY1R 3 208788_at ELOVL5 4 214306_at OPA1 5 212195_at IL6ST 6 210980_s_at ASAH1 7 204864_s_at IL6ST 8 221590_s_at ALDH6A1 9 218692_at SYBU
3. The method according to claim 1, wherein, in the measuring step, an expression level of at least one gene selected from CALB2, ASAP2, RAB5C, PTP4A2, ABCC10, OXTR, HSPA2, SMURF2, SLC7A8, RALA, ADRA2A, MYCBP, CX3CR1, ERCC1, DNAJA3, NINJ1, C4orf43, IFI35, ZNF688, SNX1, CREBL2, HPN, NME3, STS, KLF7, PDHB, NKX3-1, DEXI, GSTM3 and LCMT1 is further measured.
4. The method according to claim 3, wherein, in the measuring step, the expression level of gene is measured using probe sets specified by probe set IDs shown in Table 2-1 and Table 2-2.
   [Table 2] Table 2-1 Gene number Probe set ID Gene symbol 1 212196_at IL6ST 2 205440_s_at NPY1R 3 208788_at ELOVL5 4 214306_at OPA1 5 212195_at IL6ST 6 210980_s_at ASAH1 7 204864_s_at IL6ST 8 221590_s_at ALDH6A1 9 218692_at SYBU [Table 3] Table 2-2 Gene number Probe set ID Gene symbol 10 205428_s_at CALB2 11 206414_s_at ASAP2 12 201156_s_at RAB5C 13 216988_s_at PTP4A2 14 213485_s_at ABCC10 15 206825_at OXTR 16 211538_s_at HSPA2 17 205596_s_at SMURF2 18 202752_x_at SLC7A8 19 214435_x_at RALA 20 209869_at ADRA2A 21 203360_s_at MYCBP 22 205898_at CX3CR1 23 203720_s_at ERCC1 24 205963_s_at DNAJA3 25 211000_s_at IL6ST 26 203045_at NINJ1 27 218513_at C4orf43 28 209417_s_at IFI35 29 213527_s_at ZNF688 30 201716_at SNX1 31 201988_s_at CREBL2 32 204934_s_at HPN 33 204862_s_at NME3 34 203769_s_at STS 35 204334_at KLF7 36 208911_s_at PDHB 37 204863_s_at IL6ST 38 209706_at NKX3-1 39 203733_at DEXI 40 213702_x_at ASAH1 41 202554_s_at GSTM3 42 221515_s_at LCMT1
5. The method according to any one of claims 1 to 4, wherein the information on prognosis of breast cancer is information on risk of recurrence after surgery.
6. The method according to claim 5, wherein the information on risk of recurrence after surgery is information on risk of late recurrence after surgery.
7. The method according to claim 6, wherein the information on risk of late recurrence is information on risk of recurrence more than 5 years within 10 years after surgery.
8. The method according to any one of claims 1 to 7, wherein, in the analyzing step, the expression level is analyzed using a classification method, a hierarchical cluster analysis or a scoring method.
9. The method according to claim 8, wherein the classification method is a diagonal linear discriminant analysis or between-group analysis.
10. The method according to any one of claims 1 to 8, wherein, in the analyzing step, using the expression level and a discriminant represented by the following formula (I), solution D of the discriminant is determined, and
    in the acquiring step, risk of late recurrence of the subject is determined to be high when the solution D is a positive value, and risk of late recurrence of the subject is determined to be low when the solution D is zero or a negative value. $$\mathrm{D}={\mathrm{\Sigma }}_{\mathrm{i}}\left({\mathrm{w}}_{\mathrm{i}}\times {\mathrm{y}}_{\mathrm{i}}\right)-\left(-\mathrm{4.763756453}\right)$$

    

    wherein i represents a number assigned to each gene shown in Table 3-1 and Table 3-2, w_i represents a weighting factor of gene numbered i shown in Table 3-1 and Table 3-2, y_i represents a standardized expression level of gene according to a formula represented by formula (II): $${\mathrm{y}}_{\mathrm{i}}={\mathrm{x}}_{\mathrm{i}}-{\mathrm{m}}_{\mathrm{i}}$$

    

    wherein x_i represents an expression level of gene numbered i shown in Table 3-1 and Table 3-2, and m_i represents a mean value of expression levels of genes numbered i shown in Table 3-1 and Table 3-2 over specimens,
    and ∑_i represents a sum total over the genes.
    [Table 4] Table 3-1 Gene number Probe set ID Gene symbol Weighting factor 1 212196_at IL6ST 1.43235997 2 205440_s_at NPY1R 0.450736337 3 208788_at ELOVL5 1.793041948 4 214306_at OPA1 -2.545950461 5 212195_at IL6ST 1.539087017 6 210980_s_at ASAH1 1.539921375 7 204864_s_at IL6ST 1.151743979 8 221590_s_at ALDH6A1 1.837908119 9 218692_at SYBU 1.149172541 [Table 5] Table 3-2 Gene number Probe set ID Gene symbol Weighting factor 10 205428_s_at CALB2 -1.179415269 11 206414_s_at ASAP2 -2.323532179 12 201156_s_at RAB5C 2.316847761 13 216988_s_at PTP4A2 2.002072892 14 213485_s_at ABCC10 -2.306701292 15 206825_at OXTR -1.1298499 16 211538_s_at HSPA2 0.964294299 17 205596_s_at SMURF2 -2.109236485 18 202752_x_at SLC7A8 1.363057605 19 214435_x_at RALA -2.288374092 20 209869_at ADRA2A 1.377444437 21 203360_s_at MYCBP 1.688897659 22 205898_at CX3CR1 0.740592577 23 203720_s_at ERCC1 1.658860215 24 205963_s_at DNAJA3 2.534304027 25 211000_s_at IL6ST 0.971312607 26 203045_at NINJ1 1.889380552 27 218513_at C4orf43 1.429869188 28 209417_s_at IFI35 1.323441517 29 213527_s_at ZNF688 2.158913805 30 201716_at SNX1 2.2492475 31 201988_s_at CREBL2 1.921069642 32 204934_s_at HPN 1.115597062 33 204862_s_at NME3 1.661682832 34 203769_s_at STS -1.392023997 35 204334_at KLF7 -1.662199312 36 208911_s_at PDHB 2.071027409 37 204863_s_at IL6ST 0.825924727 38 209706_at NKX3-1 0.611126615 39 203733_at DEXI 1.847912036 40 213702_x_at ASAH1 1.575733773 41 202554_s_at GSTM3 0.648154701 42 221515_s_at LCMT1 1.7259299
11. The method according to any one of claims 1 to 10, wherein the subject is a patient with estrogen receptor (ER)-positive breast cancer.
12. The method according to any one of claims 1 to 11, wherein the subject is a subject who did not have recurrence of breast cancer for 5 years after surgery.
13. The method according to any one of claims 1 to 12, wherein the subject is a subject whose risk of early recurrence of breast cancer has been determined to be low.
14. The method according to claim 13, wherein the subject is a subject whose risk of early recurrence of breast cancer has been determined to be low, based on gene expression levels measured using probe sets specified by 95 probe set IDs shown in Table 4-1 and Table 4-2.
    [Table 6] Table 4-1 Number Probe set ID Gene symbol UniGene ID GenBank accession number 1 219306_at KIF15 Hs.646856 NM_020242 2 218585_s_at DTL Hs.656473 NM_016448 3 221677_s_at DONSON Hs.436341 AF232674 4 201088_at KPNA2 Hs.594238 NM_002266 5 209034_at PNRC1 Hs.75969 AF279899 6 202610_s_at MED 14 Hs.407604 AF135802 7 218906_x_at KLC2 Hs.280792 NM_022822 8 212723_at JMJD6 Hs.514505 AK021780 9 222231_s_at LRRC59 Hs.370927 AK025328 10 208838_at CAND1 Hs.546407 AB020636 11 218039_at NUSAP1 Hs.615092 NM_016359 12 209472_at CCBL2 Hs.481898 BC000819 13 212898_at KIAA0406 Hs.655481 AB007866 14 202620_s_at PLOD2 Hs.477866 NM_000935 15 201059_at CTTN Hs.596164 NM_005231 16 201841_s_at HSPB1 Hs.520973 NM_001540 17 203755_at BUB1B Hs.631699 NM_001211 18 211750_x_at TUBA1C Hs.719091 BC005946 19 38158_at ESPL1 Hs.153479 D79987 20 204709_s_at KIF23 Hs.270845 NM_004856 21 201589_at SMC1A Hs.211602 D80000 22 218460_at HEATR2 Hs.535896 NM_017802 23 207430_s_at MSMB Hs.255462 NM_002443 24 212139_at GCN1L1 Hs.298716 D86973 25 211596_s_at LRIG1 Hs.518055 AB050468 26 212160_at XPOT Hs.85951 AI984005 27 219238_at PIGV Hs.259605 NM_017837 28 203432_at TMPO Hs.11355 AW272611 29 201377_at UBAP2L Hs.490551 NM_ 014847 30 218875_s_at FBXO5 Hs.520506 NM_012177 31 221922_at GPSM2 Hs.584901 AW195581 32 218727_at SLC38A7 Hs.10499 NM_ 018231 33 207469_s_at PIR Hs.495728 NM_ 003662 34 218483_s_at C11orf60 Hs.533738 NM_ 020153 35 204641_at NEK2 Hs.153704 NM_ 002497 36 219502_at NEIL3 Hs.405467 NM_ 018248 37 209054_s_at WHSC1 Hs.113876 AF083389 38 220318_at EPN3 Hs.670090 NM_ 017957 39 210297_s_at MSMB Hs.255462 U22178 40 209186_at ATP2A2 Hs.506759 M23114 41 219787_s_at ECT2 Hs.518299 NM_018098 42 45633_at GINS3 Hs.47125 AI421812 43 200848_at AHCYL1 Hs.705418 AA479488 44 200822_x_at TPI1 Hs.524219 NM_000365 45 211072_x_at TUBA1B Hs.719075 BC006481 46 200811_at CIRBP Hs.634522 NM_001280 47 202864_s_at SP100 Hs.369056 NM_003113 48 202154_x_at TUBB3 Hs.511743 NM_006086 49 213152_s_at SFRS2B Hs.476680 AI343248 50 209368_at EPHX2 Hs.212088 AF233336 [Table 7] Table 4-2 Number Probe set ID Gene symbol UniGene ID GenBank accession number 51 211058_x_at TUBA1B Hs.719075 BC006379 52 209251_x_at TUBA1C Hs.719091 BC004949 53 213646_x_at TUBA1B Hs.719075 BE300252 54 204540_at EEF1A2 Hs.433839 NM_001958 55 202026_at SDHD Hs.719164 NM_003002 56 201090_x_at TUBA1B Hs.719075 NM_006082 57 213119_at SLC36A1 Hs.269004 AW058600 58 217840_at DDX41 Hs.484288 NM_016222 59 206559_x_at EEF1A1 --- NM_001403 60 202066_at PPFIA1 Hs.530749 AA195259 61 203108_at GPRC5A Hs.631733 NM_003979 62 218697_at NCKIPSD Hs.655006 NM_016453 63 222039_at KIF18B Hs.135094 AA292789 64 202069_s_at IDH3A Hs.591110 AI826060 65 203362_s_at MAD2L1 Hs.591697 NM_002358 66 202666_s_at ACTL6A Hs.435326 NM_004301 67 204892_x_at EEF1A1 Hs.520703 NM_001402 68 205682_x_at APOM Hs.534468 NM_019101 69 209714_s_at CDKN3 Hs.84113 AF213033 70 218381_s_at U2AF2 Hs.528007 NM_007279 71 201947_s_at CCT2 Hs. 189772 NM_006431 72 212722_s_at JMJD6 Hs.514505 AK021780 73 204825_at MELK Hs.184339 NM_014791 74 203184_at FBN2 Hs.519294 NM_001999 75 201266_at TXNRD1 Hs.708065 NM_003330 76 202969_at DYRK2 Hs.173135 AI216690 77 204817_at ESPL1 Hs.153479 NM_012291 78 209523_at TAF2 Hs.122752 AK001618 79 218491_s_at THYN1 Hs.13645 NM_014174 80 217363_x_at --- --- AL031313 81 218009_s_at PRC1 Hs.567385 NM_003981 82 204026_s_at ZWINT Hs.591363 NM_007057 83 218355_at KIF4A Hs.648326 NM_012310 84 202153_s_at NUP62 Hs.574492 NM_016553 85 213011_s_at TPI1 Hs.524219 BF116254 86 217966_s_at FAM129A Hs.518662 NM_022083 87 214782_at CTTN Hs.596164 AU155105 88 217967_s_at FAM129A Hs.518662 AF288391 89 204649_at TROAP Hs.524399 NM_005480 90 35671_at GTF3C1 Hs.371718 U02619 91 213502_x_at LOC91316 Hs.148656 AA398569 92 221285_at ST8SIA2 Hs.302341 NM_006011 93 221519_at FBXW4 Hs.500822 AF281859 94 202551_s_at CRIM1 Hs.699247 BG546884 95 217138_ x_ at IGL@ Hs.449585 AJ249377
15. A device for determining prognosis of breast cancer, the device comprising
    a processor, and a computer containing a memory under control of the processor,
    wherein the memory is recorded with a computer program for causing the computer to execute the steps of:

    acquiring information on expression levels of genes of IL6ST, NPY1R, ELOVL5, OPA1, ASAH1, ALDH6A1 and SYBU in an RNA sample prepared from a biological sample collected from a subject;

    determining prognosis of breast cancer based on the information on the expression levels of genes; and

    outputting a determination result.

Images